Behaviour of columns in sub-frames with semi-rigid joints
The behaviour of limited subassemblages with flexible beans and semi-rigid beam to column connections was studied using a computer program in which the finite element method was employed in a non-linear analysis which accounts for the presence of semi-rigid connections and the inelastic behaviour of frames. The program accounts for many other factors such as initial imperfections and residual stresses. The theoretical background to the present computer program has been presented along with the program layout. The program was used to simulate some of the experimental results obtained from tests on rigidly and flexibly connected frames with different combinations of beam and column loads. The analytical results were found to compare reasonably well with the experimental results. The program was also used to simulate a series of I-shaped subassemblages that were tested at the University of Sheffield. Comparisons were made between the analytical and experimental results characterized by the maximum loads, load-deflection curves and load- moment curves. Good agreement was obtained between the analytical and the experimental load-deflection curves for all of the cases considered. The general trends of the measured and calculated load- moment curves for most cases were found to be comparable. The recommendations given in B35950 for the design of columns in simple construction were applied to all cases in the last series and were found unconservative in the cases of balanced loading and conservative in the cases of unbalanced loads. A limited parametric study was conducted to study the effects of semi-rigid joints, bean flexibility and type of loading. In this study, an I-shaped subassemblage was analysed for different load types and different types of beam to column connections. A substantial effect was recognized due to the presence of semi-rigid connections whether or not a beam load was applied. Beam flexibility was also seen to affect the carrying capacity of the subassemblage under the action of column load only although this effect was less noticeable than that of the connection flexibility. The presence of beam load was found to result in an unexpected interaction curve which relates the total force in the column to the moment that is transmitted to the column's end. An almost linear relationship with negative gradient seems to exist between the column and beam loads. It is pointed out that all the findings of the present study are based on the range of cases considered in the parametric study but it is suggested that they serve as indicators to the behaviour of any the subassemblage under axial load oniy or axial load combined with beam loads. A few recommendations for future work are presented.